JP2004120108A - Image supervisory system and image supervisory system employing network - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image supervisory system which immediately and accurately grasps a distinct profile unique to a person committing a criminal or illegitimate act in a shop to create a likeness and automatically informing a shop clerk or an external agency or the like about it. <P>SOLUTION: The image supervisory system is provided with: many imaging means installed at prescribed points at which people enter and leave an entrance; an image processing means for extracting a distinct profile of a face of a person from video information imaged by the imaging means; a similarity discrimination section that converts the distinct profile of the face obtained by the image processing means into graphic data and discriminates the similarity of the graphic data to graphic data of a plurality of standard images registered in advance; a likeness generating means that combines the respective graphic data on the basis of position information of the graphic data of the standard images registered in advance when the discrimination result of the similarity discrimination section indicates a similarity; and a means for providing the graphic data and the distinct profile image of the photographed video information to a monitor requiring them. <P>COPYRIGHT: (C)2004,JPO

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to an image monitoring system that detects features of a customer who visits a supermarket, a playground, or the like, creates a portrait, and grasps a store state or a visitor's situation.
[0002]
[Prior art]
In recent years, the number of stores such as supermarkets, convenience stores, and pachinko machines has increased, and these stores tend to be long hours, such as late at night. A monitoring system has been established.
[0003]
Also, it is well known that amusement centers such as arcades and pachinko parlors have a large number of TV camera surveillance systems built to manage customer safety and monitor fraudulent activities by customers and clerks. It is.
[0004]
Particularly, in a game hall or the like, when a gangster or the like enters or exits, or when there are many customers with a bad playing attitude, the number of sensible customers coming to the store decreases, which is a major problem in store management. Therefore, an effective monitoring system is desired.
[0005]
Such a surveillance system is, of course, installed in a financial institution such as a bank and prevents crimes such as robbery, and uses the obtained images as evidence for criminal investigations.
[0006]
The images captured by the above-mentioned many television cameras are intermittently recorded for a long time by a recording device such as a time-lapse VTR. Although the recorded content of the recorded video can be checked with various data at a later date, the quality of the video is not so high at present.
[0007]
Therefore, as a system for monitoring the movement of a moving object such as a person as a security device, a method in which a monitoring camera is appropriately driven according to the reaction of an ultrasonic or infrared sensor and a captured image is recorded as it is or an image obtained by the monitoring camera Surveillance systems have been constructed by various methods of extracting and recording a motion component of an object from an object.
[0008]
There is also a publication (Patent Document 1) for creating a portrait that is closer to the real thing.
[0009]
[Patent Document 1]
JP-A-6-205353
[0010]
[Problems to be solved by the invention]
However, these conventional surveillance systems always store and record a captured image obtained by driving a camera and analyze it at a later date to know that the person is a criminal offender or a criminal act in comparison with a facial photograph or the like. Since the system is a system, there is a problem that even if the entrance is monitored by a surveillance camera, it is not possible to immediately detect that a person needs attention or a criminal when entering the store.
[0011]
There is also a method of transmitting a photographed image of a person of interest to another store or a monitor, but there is a problem that the cost is high because the amount of data is large. Intuitively, caricatures are sometimes better.
[0012]
Therefore, it becomes expensive, and it is difficult to prevent crime or wrongdoing in the store beforehand because it is not possible to intuitively and appropriately judge the person to be watched by looking at the image (photograph). Met.
[0013]
Also, as can be seen from the fact that caricatures are often used in police open investigations, etc., caricatures are easier to identify than pictures when specifying a person's face.
[0014]
This is because it is difficult to specify a person in a photograph that is described in detail in detail, because a lot of information other than the feature rather than makes the impression lighter.
[0015]
However, there is a case where the culprit cannot be easily identified only by the picture.
[0016]
The present invention has been made in order to solve such a problem, and immediately and accurately grasps the unique characteristics of a person who commits a crime or misconduct in a store, creates a portrait, and automatically notifies a salesclerk, an external organization, etc. The purpose is to obtain an image surveillance system that can.
[0017]
[Means for Solving the Problems]
An image monitoring system according to the present invention includes a plurality of image pickup units installed at predetermined locations where a person entering and exiting an entrance can be photographed, an image processing unit that extracts a facial feature from video information imaged by the image pickup unit, A similarity determination unit that converts the feature of the face obtained by the image processing unit into graphic data and determines similarity with graphic data of a plurality of standard images registered in advance, and a determination result of the similarity determination unit is When indicating similarity, a portrait creating means for combining the respective graphic data based on the position information of the graphic data of the pre-registered standard image, and a feature image of the graphic data and the photographed video information. And a means for providing the necessary monitor.
[0018]
An image monitoring system using a network according to the present invention is an image monitoring system using a network for connecting a monitoring system of each building and a central management center to a computer network and exchanging information of persons with each other.
[0019]
The monitoring system of each building is provided with a plurality of imaging means at a predetermined location where a person entering the building can be photographed from the front, left and right directions of a gate entering the building, and extracting a characteristic image of a predetermined part of the person from the video information, The gist of the invention is to provide a means for generating graphic data of a person of interest, and a means for transmitting the graphic data to a monitoring system of a different building, and then transmitting an image captured by the imaging means as required.
[0020]
Further, an image monitoring system using a network according to the present invention is an image monitoring system using a network that connects a monitoring system of each building and a central management center to a computer network and exchanges information of persons with each other.
[0021]
The central management center collects video data of the criminal from the criminal information providing center, and extracts the characteristics of the person image from the video information collected by the collection unit, and creates graphic data of a predetermined part. Means, a similarity determination unit that determines the graphic data from the monitoring center by comparing graphics of a plurality of standard images registered in advance, and when the determination result of the similarity determination unit is determined to be similar, Means for providing the collected video data to the monitoring system.
[0022]
BEST MODE FOR CARRYING OUT THE INVENTION
<Embodiment 1>
An outline of the image monitoring system according to the first embodiment of the present invention will be sequentially described with reference to FIG. In the present embodiment, as an example of a store to which the image monitoring system is applied, a pachinko parlor, which is a playground, will be described as an example.
[0023]
FIG. 1 is a configuration diagram of an image monitoring system according to an embodiment of the present invention. This image monitoring system includes a video camera 10a, 10b, 10c for photographing an image of a person at the entrance, a television camera 11a ... 11j for photographing a pachinko stand passage in a store, and a television camera 12a ... , A variety of sensors 23 (23a, 23b...) Built in a large number of amusement tables 20 (20a, 20b...), 21 (21a, 21b...), And 22 (22a, 22b. The small CCD cameras 26 (26a, 26b ...), 27 (27a, 27b ...), 28 (28a, 28b ...) built in the gaming table are connected to the office system. A clerk in the store carries an intercom slave unit 24.
[0024]
Three video cameras 10a, 10b, and 10c among the above-described video cameras are arranged at appropriate angular positions on the front and both sides of the entrance, respectively, so that the characteristics of a person entering the store can be photographed. As a result, the color of the clothes as well as the features of the occupant's face can be acquired as data at the same time.
[0025]
Further, other video cameras are installed at high places such as ceilings in the store so that the person to be monitored and the operation state can be easily monitored toward each passage in the store.
[0026]
These video cameras have a zoom-in function as well as a pan and tilt function (a video camera platform).
[0027]
On the other hand, in the office, a controller 30 that controls the video cameras 10 to 12, a VTR 31 (or a disc recorder) that records images of the video cameras 10 to 12 based on the control of the controller 30, For example, a system controller 33 that controls a number of pachinko machines based on detection signals from various sensors 23 attached to the devices 21 and 22 and a video signal from a number of video cameras input via the controller 30 An image processing device 40 such as a DSP is provided.
[0028]
Further, based on the image processed by the image processing unit 40, characteristic image data such as a person's face and clothing color is acquired, and the acquired characteristic image data is compared with the registered image data. In addition to notifying the result to the unit 33, the characteristics of each part of the person's face are obtained, and the characteristic image of each part is converted into graphic data Gi, and the graphic data gi of each part stored in the main storage device 60 in advance is stored. (A stored portrait data), an image determination / management device 50 for notifying the result of the comparison, an image of a person of interest, a table of characteristics of a person of interest, a wanted criminal, etc. A main storage device 60 including, for example, a hard disk which stores a table and the like and stores image data of the occupant acquired by the image determination / management unit 50; A public address system 70 connected to the scan 90, and a monitor 80. for displaying the state of the store on the basis of a video signal from the video camera.
[0029]
The main storage device 60 stores graphic data gi of each part (face outline, eyes, nose, ears, eyebrows ...) of the attention person in position information (eye: distance between eyes, ear: mm from mouth). ..) Are stored.
[0030]
The various sensors provided in each of the amusement tables 20, 21, 22 include, for example, a ball detection sensor, a glass breakage detection sensor, a door open detection sensor, a magnet detection sensor, a cell detection sensor, a ROM detection sensor, and the like. Note that the detection items of these sensors are stored in the main storage device 60 in a table in advance. This is data for the system control unit 33 to recognize which sensor has been activated from the input detection signal.
[0031]
The controller 30 zooms in or out of the video camera based on a control signal from the system control unit 33.
[0032]
When the controller 30 is notified by the system control unit 33 that any one of the various sensors of each game console has detected, image data from a predetermined video camera using the game console as a monitoring area. Is recorded on the VTR 31.
[0033]
At this time, the image detected by the sensor is switched and photographed prior to the normal monitoring image, and is displayed on the monitor 80.
[0034]
Further, a pachinko machine camera controller 35 is provided which drives the small CCD cameras 26, 27, 28 built in each pachinko machine, stores the obtained video signal in the VTR 36, and sends it to the image determination / management device 50. I have.
[0035]
As shown in FIG. 2, the image determination / management unit 50 includes a face detection unit 50a, a face direction detection unit 50b, a face feature detection unit 50c, a clothing detection unit 50d, a similarity determination unit 50e, and a portrait drawing unit 50b. , A report determination unit 50e, a RAM 50f for temporarily storing various image data acquired by these units, and a CPU 50g for controlling the entire image processing unit 2.
[0036]
The face detection unit 50a detects the contour of the face from the video signals of the video cameras 10, 11, and 12 input via the image processing unit 40, and attaches an ID for identifying the person having the face. When detecting the outline of a face, whether or not the face is a face is determined by skin tone (skin color detection).
[0037]
The face direction detection unit 50b determines whether the face is “front”, “right side”, “left side” or “rear” when the face is graphicized by the portrait creation unit detected by the face detection unit 50a. Judgment is made from the face orientation table (b) shown in FIG. 3 stored in the main storage device 60 as a database, and the outline of the face is extracted and the outline of the face as an external feature is set to “circle”, “triangle”, “ It is determined from the contour table (c) whether it is an inverted triangle or a square or the like, and the presence or absence and the shape of the glasses are selected from the glasses table (d).
[0038]
The face feature detection unit 50c determines whether or not “eye”, “nose”, “eyebrows”, “mouth”, “ear”, “whisker”, and “(face part feature table) hair” exist in the acquired face information. Features such as (face feature information Ri) are extracted, and respective coordinates are calculated and stored in association with each other. In addition, the portrait creating unit 50h converts the facial feature information Ri of “eyes, nose, eyebrows, etc.” acquired by the facial feature detecting unit 50c into a graphic, and stores it as portrait data Gi.
[0039]
The similarity determination unit 50e determines the similarity with the face part feature Hi registered in the main storage device 60 in advance based on the face feature information Ri. At this time, since the glasses can be easily attached and detached and replaced, they may not be used for similarity determination.
[0040]
Further, the similarity between the portrait data Gi of each part created by the portrait creation unit 50h and the portrait data gi in the main storage device 60 is determined (the presence or absence of a match).
[0041]
The report determination unit 50e determines whether or not several pieces (eyes, ears, and nose) of the face feature information Qi obtained by the face feature detection unit 50c match the pattern of the face part feature data Hi, or a portrait. When several portraits (eye, ear, nose) of the portrait data Gi created by the creation unit 50 match the portrait data Gi, the created portrait data Gi and the acquired image data Qi are transmitted to the intercom system and the intercom slave unit. I do.
[0042]
For example, the similarity determination unit 50e stores the face part feature data Hi of regular customers previously stored in the main storage device 60 in advance in the main storage device 60, a group of customers who have committed misconduct in the past, and a group referred to as a master. Is compared with the facial part feature data Hi. In addition, face part feature data Hi of regular customers previously stored in the main storage device 60 in advance, and face part feature data Hi of a group that is referred to as a customer who has committed misconduct in the past and a master. Compare with However, the facial features are changed to portrait data Gi. The result is notified to the system control unit 33.
[0043]
In addition, a feature table is created from images of important wanted persons and the like to be opened to the public so that they can cooperate with criminal investigations, and is used for comparison.
[0044]
The facial part feature data Hi and the portrait data Gi of these persons for which the similarity determination is required are stored in the main storage device 60 as described above, and the text data ( Portrait).
[0045]
The portrait data gi such as “eye”, “nose”, and “ear” shown in data (c), (e) to (i) of FIG. 3 is stored in the main storage device 60 as described above.
[0046]
The portrait creating section 50h uses the coordinates of the facial feature information Ri such as “eye”, “nose”, and “ear” detected by the facial feature detecting section 50c to obtain the data (c), (e) to (e) in FIG. (I) is selected and combined to create a portrait ZG.
[0047]
The on-site broadcasting system 70 is, as shown in FIG. 4, a voice synthesizing unit 71 for inputting voice data read from the main storage device 60 under the control of the system control device 33 via the system bus 90, and a main control unit. An intercom system control unit 72, a plurality of master units 73 which are relay stations connected by a wire to the system control unit 72 (one master unit is sufficient for a small shop), and a wired connection to the intercom system control unit 72 And a plurality of slave units 76 which are terminal stations capable of transmitting and receiving information to and from the master unit 73.
[0048]
The above-described intercom system control unit 72 converts the audio data input through the audio synthesis unit 71, for example, “Thank you for visiting every time” into a digital signal by the A / D converter 72a, and performs the audio MIX & communication path. After the signal is processed by the A / D converter 72c via the switch 72b, the signal is output to the speaker 74 in the hall. At this time, when sound is input from the microphone 75, the sound is mixed by the sound MIX & communication path switch 72 b and output to the speaker 74 as a mixed signal.
[0049]
Further, for example, in-flight broadcasting from the microphone 75 and tone information from a music reproducing device (not shown) can also be output as BGM from the in-floor speaker 74. As the tone information, for example, MIDI information is adopted and stored in the main storage device 3.
[0050]
Such music information can be managed so as to improve the customer's motivation by, for example, playing fast-paced music to a group of amusement tables or rows in a row with poor pitching. The appropriate program system is appropriately performed under the control of the system control unit 33 which manages payout information and the like.
[0051]
The intercom system control unit 72 includes an I / F (interface) 72d connected to a public line, and an A / D converter 72e inserted between the I / F 72d and the audio MIX & communication path switch 72b. , An external I / F (external interface) 72f connected to the plurality of master units 73.
[0052]
The master unit 73 includes an I / F 73a connected to an external I / F 72f of the intercom system control unit 72, a modem 73b, and an R / F circuit 73c for wireless transmission and reception between the plurality of slave units 76. . Each of the slave units 73 is owned by a store clerk, and communicates in-store information with the master unit 73 as appropriate, exchanges between a number of slave units 76, and security communication modes only between the slave units 76 in a group. Are set and used as needed.
[0053]
Preferably, the terminal possessed by the clerk is capable of displaying a screen such as an i-mode (registered trademark) mobile phone.
[0054]
The overall operation of the video surveillance system configured as described above will be described below with reference to the sequence diagram of FIG. In this description, an image obtained by a video camera at the entrance and an image obtained by a small CCD camera of a pachinko machine will be described as examples.
[0055]
For example, when a person enters from the entrance of a pachinko parlor, the person is photographed by the video cameras 10a, 10b, and 10c from the front and left and right. The front image and the left and right images are input to the controller 30 (d1, d2, d3).
[0056]
The controller 30 records the front image and the left and right images on the VTR 31 and sends them to the monitor 80 and the image processing device 40 (d7, d8, d9). At this time, the shooting time is also transmitted.
[0057]
The image processing device 40 receives the front image and the left and right images, performs back surface removal processing on each image, and extracts a person image from the front and a person image when photographed from the left and right, An ID with which each is associated is added.
[0058]
For example, IDs of IDA1 are added to the person image from the front, and IDA2 and IDA3 are added to the left and right person images.
[0059]
Then, the image processing device 40 sends the front person image (IDA1), the right person image (ID2), and the left person image (ID3) to the image determination / management device 50 with the shooting time as the acquired image data. It is sent (d10).
[0060]
The image determination / management device 50 receives the acquired image data Qi and temporarily stores it in the database 60 (main storage device) (d11).
[0061]
Then, contour extraction (face feature information Ri (face contour, nose, eyes)) is performed in the acquired image Qi, and the attention person pattern hi and the attention person portrait data gi stored in the database 60 in advance are used. Is used to determine the characteristics of the person.
[0062]
For example, the height of the person is measured using the outline of the person image in which the entire outline appears in the acquired image data Qi. Hereinafter, in this example, comparison of portrait data Gi will be described.
[0063]
Then, it is determined which height the height belongs to in the height file of FIG. 3A (the height determined to belong is gai). Next, it is determined to which contour the contour of the front person image belongs in the contour file of FIG. 3C (the contour determined to belong to gci).
[0064]
Further, it is determined to which glasses in FIG. 3D the glasses of the front person image belong (glasses determined to belong to gdi).
[0065]
Next, it is determined to which eye in FIG. 3E the eye of the front person image belongs (the eye determined to belong to gei).
[0066]
Next, it is determined whether the nose of the front person image belongs to which nose in FIG. 3F (the nose determined to belong to gfi).
[0067]
Further, it is determined to which ear in FIG. 3 (h) the ears of the left and right person images belong (ears determined to belong to ghi). Further, it is determined to which of the mustaches in FIG. 3 (i) the beard of the front person image belongs (the mustache determined to belong to gii).
[0068]
Further, it is determined to which eyebrow the eyebrows belong in FIG. 3 (g) (the eyebrows determined to belong to gfi).
[0069]
In this way, the facial feature information Ri of the person entering the store from the entrance is recognized graphically (d12). That is, in this example, as shown in FIGS. 3 and 6, the person who enters the store has a height of 171 to 180 (gai), a face outline of a square (gci), eyes of eyes (gei), and a nose of a lion nose ( This means that the person is recognized as having a characteristic of gfi).
[0070]
Then, a feature file GTi (not shown) to which the portrait data Gai, Gci, Gdi,... Of the determined attention person is allocated is generated and stored in the main storage device 60 together with the already acquired ID. At this time, the position (eye position, nose position,...) Is stored in correspondence with each person Gai,.
[0071]
Next, the portrait drawing section 50h renders each part of the facial feature information Ri into a graphic Gi (d13).
[0072]
Then, the similarity determination section compares the graphicized face outline graphics Gai,... With the attention person graphic gi stored in advance in the database 60, and determines whether or not there are several matching ones (d15).
[0073]
In addition, face part feature information Ri of each person is stored in the database 60. The face part feature data Hi and hi of the attention person are compared to determine whether or not there is a match (d14).
[0074]
Then, when Hi = Ri or Gi = gi, the notification determination unit 50e notifies the intercom system of a caution (d16), assembles the graphic data Gi based on the position coordinates, and transmits this to the monitor (d17). At the same time, the acquired image Qi is transmitted to the monitor (d18).
[0075]
That is, as shown in FIG. 7, the clerk can easily find the attention person because the graphic (assembled portrait GM) and the raw image of the attention person are displayed on the monitor.
[0076]
The image transmission of the attention person is performed by a public investigation stored in this table and database in advance, so that it is possible to cooperate with criminal investigations as well as the characteristics of persons who have been cheating in the past and who are said to be masters. It is determined from the image of the important wanted person or the like or whether or not it is similar to the important wanted person.
[0077]
At this time, since the glasses can be easily attached / detached / replaced, gdi in (d) is not used for similarity determination.
Further, for example, when the similarity is determined to be very similar, the inker system plays characteristic music from a speaker (d19). That is, the clerk is notified by music or the like that the person to be careful is visiting. If the music is not automatically switched, the clerk uses a microphone to notify the attendant of the person in need of attention from the speaker using a memo or the like.
[0078]
If the clerk has a terminal capable of displaying a screen, the portrait is also sent to the clerk's terminal and displayed. In addition, a message that a cautionary person has entered the store is notified to the child device of the clerk.
[0079]
At the same time, the criminal information is displayed on the monitor, for example, what kind of wrongdoing has been done in the past, who is the master, and what kind of criminal is an important wanted person.
[0080]
On the other hand, the system control device 33 reads detection signals from various sensors (such as detection of a ball coming out, glass breakage detection, door opening detection, magnet detection, cell detection, ROM detection, etc.) attached to each pachinko machine (d19). If the detection signal indicates an abnormality, the number of the pachinko machine is read, and the CCD camera 26 of this pachinko machine is operated (d20). The image captured by the CCD camera 26 is sent to the image processing device 40 via the controller 35 (d21).
[0081]
In the image processing apparatus 40, the back surface removal processing is performed, and the image of the person sitting on the table is extracted and an ID is added.
[0082]
If it is not possible to make an accurate determination based on the image taken by the camera installed on the table, the comparison can be made with reference to the facial features captured when entering the store.
[0083]
The image of this person is sent to the image determination management device (d22), stored in the database (d23), and the face characteristic information Ri and the face graphic data Gi are obtained by the above processing (d24). The comparison with the person in need of attention is performed, and the result is notified to the intercom system, and the message and music are changed as described above, and the face graphic Gi and the data Qi are sent (d27) and displayed on the monitor ( d28).
[0084]
At this time, if they are not similar as a result of comparison with the person requiring caution, the person is stored in the database as a new person requiring caution.
[0085]
Further, when the detection signal notifying the abnormality is detected from the sensor, the image determination management device sends a camera number for monitoring the path where the abnormal table is arranged to the controller (d30).
[0086]
The camera sends a large number of human images of the passage to the image processing device 40 via the controller (d31).
[0087]
The image processing device 40 extracts only a large number of human images as described above and sends it to the image determination management device (d32). The image determination management device stores a large number of these person images in the passage in the database (d33), obtains the characteristics of the person sitting on the table by the above-described processing, and compares the characteristics with the person requiring attention in the database. (D34), the result is notified to the intercom system, the message and music are changed as described above, and the portrait GM, the raw image Qi, and the crime information of the person in need of attention are displayed on the monitor (d35).
[0088]
Therefore, the clerk can rush to the site according to the contents of the message and can check the surrounding state, so that it is possible to prevent fraudulent acts.
[0089]
In addition, since the portrait of the person in need of care is viewed in advance, the person in need of care can be easily found on site. In addition, since the crime information of the person requiring attention is known, it is possible to predict what kind of wrongdoing is going to be done and what kind of crime is going to be done. In the case of a violent criminal, immediately reporting to the police can protect customers and clerks from danger.
[0090]
This makes it possible for the store to cooperate with the criminal investigation, which is active as a citizen. This has the advantage of improving the atmosphere inside the listing and allowing customers to play with peace of mind.
[0091]
Furthermore, if a portrait of a wanted person or the like is displayed in advance not only on a monitor in a hidden position for the clerk but also on a monitor installed in a position that can be easily seen by customers, not only the clerk but also all customers in the store can be suspicious Can greatly improve the discovery rate and ensure the safety inside the store.
[0092]
In the above embodiment, the monitoring of a person in a pachinko parlor has been described, but the present invention may be applied to a supermarket, a convenience store, and a department store.
[0093]
<Embodiment 2>
In addition, as shown in FIG. 8, a monitoring system for each store such as a supermarket, a convenience store, and a pachinko parlor, and a centralized management center and a criminal information providing organization are connected to each other via a network such as the Internet or an intranet. Alternatively, crime information or the like may be exchanged to contribute to crime prevention.
[0094]
The video surveillance system described above is arranged in these stores, and the acquired characteristic table of the person requiring caution is transmitted to the central management center 100.
[0095]
The central management center 100 collects criminal information (face part feature data Hi, caricature data Gi) from the criminal provider organization 101, compares them with each other for each predetermined part, determines the same or similar, and determines a feature table (caricature). ) Is created and depends on the feature table of the person requiring attention received from each store.
[0096]
That is, in the first embodiment, the characteristic table of the person requiring attention and the characteristic table of the criminal are stored in the database of his / her store. In the second embodiment, the characteristic table is compared with the information via the network 102. The results are obtained and reported to the clerk by the monitor and the income system.
[0097]
For example, as shown in FIG. 9, when the image monitoring system of the pachinko machine A detects an abnormality (the system controller determines the abnormality from the detection signal from the sensor of the pachinko machine) (d40), the image is obtained by the CCD camera 26 of the pachinko machine. The image determination / management apparatus 50 transmits the portrait GM of the person assembled on the basis of the portrait image, the portrait data gi combined at this time, and the shop number mi to the central management center 100 (d41).
[0098]
A computer (not shown) of the central management center 100 determines similarity from the portrait data Gi (person who is a goto, a fraudulent person, and a criminal (generically also referred to as a person requiring caution) of a person, and similar data exists. If so, it is determined that there is a registrant (d42), and the criminal information H (including the portrait data Gi) of the person requiring caution is transmitted to the pachinko parlor A via the network 102 (d43).
[0099]
When the image determination / management device 50 of the pachinko parlor A is transmitted from the computer of the central management center 100 that there is a registrant, the image is notified to the clerk by using the internal broadcasting system 70 and the type of music is changed.
[0100]
At the same time, a portrait is created from the generated feature table (Hi) and the standard image (FIG. 3) stored in the database 60 in advance.
[0101]
Then, the monitor in the store displays a portrait of a cheating person or the like.
[0102]
If the clerk has a terminal capable of displaying a screen, the portrait is also sent to the clerk's terminal and displayed.
[0103]
When the central management center 100 determines from the pachinko parlor A that the registrant does not exist in each file from the person characteristic table gi (d44), the portrait data Gi is registered in the unauthorized person file as a new person requiring caution. (D47). At this time, the central management center 100 transmits the raw data Qi from the store A, and generates portrait data from this (d47a, d47b).
[0104]
When the image determination / management device 50 of the pachinko parlor A determines that the person requires caution, the store number mi and Gi and gi of the cautionary person are connected via a network to a nearby pachinko parlor (for example, in the same area). (Store B with URL code of store A) (d48).
[0105]
The store B in the same area to which the characteristics Gi and gi of the person requiring attention are transmitted creates a portrait from the transmitted characteristic table and the standard image stored in the database of the store B. The store B displays the portrait data Gi on the monitor, inputs the URL of the screen (shown in FIG. 10) as necessary, and receives the raw data Qi from the store A (d49, d50). The created portrait is displayed on a monitor in the store.
[0106]
That is, for example, when a goto master appears in pachinko parlor A, a portrait is displayed immediately at, for example, pachinko parlor B, so that the pachinko parlor B can take measures in advance.
[0107]
Since only the portrait is sent from the central management center to the store B without sending the image file, the data amount is small, and a broadband communication line is not required. That is, if a modem is used on a normal telephone line, it can be used without any construction.
[0108]
The video camera according to the above-described embodiment is attached to a ticket gate of a station as shown in FIG. 11, and the image is transmitted to a site (not shown) via a network. It may be determined whether or not the person is the person, and the result may be reported to the police. It is desirable that this report information be reported with the station name added.
[0109]
Further, the display device of the above-described embodiment may be installed in a place such as a station yard or a bus terminal where a fleeing wanted person can pass, and a portrait of the wanted person may be displayed. In this case, many wanted persons can be displayed by sequentially switching the portrait screen.
[0110]
In addition, since a caricature can be displayed immediately after a crime occurs in a store or the like, a caricature can be immediately displayed at a station or the like while the criminal is escaping from the crime scene, and quick discovery can be performed.
[0111]
【The invention's effect】
As described above, according to the image monitoring system of the present invention, when a plurality of imaging means arranged at a predetermined location shoots, a person image is extracted from this image to determine the characteristics of each part of the person image and generate graphic data. I do.
[0112]
Then, as a result of comparing the graphic data with the graphic data registered in advance, the portrait and the raw data of the graphic data are displayed on various notification devices in the store.
[0113]
For this reason, the clerk in the store can easily recognize the presence of a person requiring attention from the graphic and confirm it with raw data (camera image), so that it is possible to prevent criminal acts beforehand, and a portrait (Graphic) makes it easy to identify the person in need of attention.
[0114]
For example, at a pachinko parlor, it can be seen that Goto came to the store. In addition, in the supermarket, it can be seen that shoplifting addicted criminals visited the store. And even if the person in need moves, you can easily find it with a portrait.
[0115]
Furthermore, if a portrait of a criminal targeted for a wide area investigation is always displayed, a public can cooperate in a criminal investigation at a store, a supermarket, a convenience store, a station, a bus terminal, and the like.
[0116]
In addition, when an abnormality is detected, a camera that monitors the location of the sensor that detected the abnormality is activated and an image is acquired, so it is possible to judge that there are multiple people near the playground Thus, the result of the determination can be notified to the child machine of the clerk. For this reason, fraudulent acts can be prevented.
[0117]
Further, according to the image monitoring system using the network of the present invention, after the shop that has detected the attention person sends the graphic data to another shop, the shop sends the raw data on demand.
[0118]
For this reason, there is no need to prepare a large-capacity storage device in the database, and there is no need to transmit and receive a large amount of data through a network, and a large-capacity network is not required, so that communication costs can be reduced.
[0119]
In addition, when a fraudulent activity or a criminal is detected, the central management center notifies the nearby building (store) and displays a portrait, so that measures can be taken to prevent fraudulent activity in other stores. .
[0120]
Also, if the camera of the present system is attached to the gate of the station, the wanted person can be immediately notified to the police, so that a wide area security system can be provided.
[0121]
Furthermore, when the display device of the present system is installed at a station, a portrait can be seen by a public immediately after the crime, and the culprit can be found during escape.
[Brief description of the drawings]
FIG. 1 is a schematic configuration diagram of a video monitoring system according to an embodiment.
FIG. 2 is a schematic configuration diagram of an image determination / management device.
FIG. 3 is an explanatory diagram illustrating a table structure for recognizing a person.
FIG. 4 is a schematic configuration diagram of an on-site broadcasting system.
FIG. 5 is a sequence diagram illustrating the overall operation of the present embodiment.
FIG. 6 is an explanatory diagram illustrating various parameters of the present embodiment.
FIG. 7 is an explanatory diagram of a screen displayed on the monitor according to the first embodiment.
FIG. 8 is a schematic configuration diagram of a second embodiment.
FIG. 9 is a sequence diagram schematically illustrating an overall operation of the second embodiment.
FIG. 10 is an explanatory diagram of a screen displayed on a monitor of a store B according to the second embodiment.
FIG. 11 is an explanatory diagram in which a CCD camera is provided at a ticket gate of a station.
[Explanation of symbols]
10 Video camera
20 play tables
23 Various sensors
33 System control unit
40 Image processing device
50 Image judgment and management unit
50a face detection unit
50b Face direction detection unit
50c face feature detection unit
50d Clothes detector
50e similarity determination unit

Claims (4)

A plurality of imaging means installed at a predetermined location that can photograph a person entering and exiting the entrance,
Image processing means for extracting a facial feature from the video information imaged by the imaging means;
A similarity determining unit that converts the feature of the face obtained by the image processing unit into graphic data and determines similarity with graphic data of a plurality of standard images registered in advance;
When the determination result of the similarity determination unit indicates similarity, a portrait creating unit that combines the respective graphic data based on the positional information of the graphic data of the pre-registered standard image,
Means for providing the graphic data and the characteristic image of the photographed video information to a monitor that requires it;
An image monitoring system comprising: When it is not possible to extract the data of the predetermined part showing the characteristic from the person image obtained by the one image pickup means, the characteristic image is extracted from the person image obtained by the other image pickup means, 2. The image monitoring system according to claim 1, further comprising: means for sending to the comparing means. An image monitoring system using a network that connects a monitoring system of each building and a central management center to a computer network and exchanges information of persons with each other,
The monitoring system of each building,
A plurality of image pickup means are provided at a predetermined position of a gate that enters the building, where a person can be photographed from the front, left and right directions,
Means for extracting a feature image of a predetermined part of the person from the video information and generating graphic data of the person of interest;
After transmitting the graphic data to a different building monitoring system, a means for transmitting a video imaged by the imaging means as desired,
An image monitoring system using a network, comprising: An image monitoring system using a network that connects a monitoring system of each building and a central management center to a computer network and exchanges information of persons with each other,
The central management center,
Collection means for collecting criminal video data from the criminal reporting center;
Means for extracting features of a person image from the image information collected by the collection means, and creating graphic data of a predetermined part,
A similarity determination unit that determines by comparing graphic data from the monitoring center with graphics of a plurality of standard images registered in advance;
Means for providing the video data collected in advance to the monitoring system when the determination result of the similarity determination unit is similar, the image monitoring system using a network.

Images